ErgoDox - Custom split ergo keyboard.

[laffindude]

Hi. I am still in this even if I don't really check the thread. I see we're still stuck on the case. I am gonna fire off an email to a friend to see if he can help with the aluminum case.

[tradet]

Sounds tempting. Interested depending on the price. $200 maybe?

[jblack801]

After a few tries you get the amount of solder correct. After that it is a smooth experience =)

I may be a bit of a soldering amateur, but damn that was awesome.

[sordna]

This video is very impressive, such CLEAN soldering !!!

[video=youtube;5uiroWBkdFY]http://www.youtube.com/watch?v=5uiroWBkdFY[/video]

[ic07]

Got I2C working the other day!  And spring break's coming up, so I'm hoping (God willing) to have at the very least some beta quality firmware done by the end of next week.  I'll post my documentation and code when I get it working.  Sorry it's taken so long, but thanks for being patient .

[dorkvader]

This is very good news, ic07! I am impressed with both your dedication and your results. We are patient here: Time is what is necessary for proper design.

[bpiphany]

G'damn these things are small. I ordered a bunch of 0402 components for using when reconstructing the Teensy directly on the main board. Time to do some practice I think, buying a finer tipped soldering iron, and finding someone willing to borrow me a stereo loupe perhaps...

[video=youtube;66GV4OuShzI]http://www.youtube.com/watch?v=66GV4OuShzI[/video]

[alaricljs]

Tin both pads, hold the 0402 down to the board with tweezers hit 1 pad with the soldering iron briefly.  Check to see if the other pad got sufficiently heated to be properly soldered.  Sometimes it happens.  Then finish it out.

[dirge]

keep it up guys, in for at least one regardless of price.

[QuinRiva]

I've been interested in making something along these lines for a while, and I've done a quick mock up based on Dox's design.  Mine will largely be sheetmetal based as that is the tools I have access to, so I'm looking at having the keys plate mounted and having the left and right keyboard mounted on a chair.  As far as curvature goes that isn't really much of a problem for me as I can do that pretty easily, but I've been having trouble trying to determine the best direction for curvature and the dimensions.

Also what have you used to draw the model and where have you sourced the switch models?

[ Attachment Invalid Or Does Not Exist ] 47545[/ATTACH]

[Dox]

Got I2C working the other day!  And spring break's coming up, so I'm hoping (God willing) to have at the very least some beta quality firmware done by the end of next week.  I'll post my documentation and code when I get it working.  Sorry it's taken so long, but thanks for being patient .

Great! I'll wait for your documentation!

G'damn these things are small. I ordered a bunch of 0402 components for using when reconstructing the Teensy directly on the main board. Time to do some practice I think, buying a finer tipped soldering iron, and finding someone willing to borrow me a stereo loupe perhaps...

[video=youtube;66GV4OuShzI]http://www.youtube.com/watch?v=66GV4OuShzI[/video]

Man, thee things are sooo tiny, you should have gone with some 0603 or 0805.

I've been interested in making something along these lines for a while, and I've done a quick mock up based on Dox's design.  Mine will largely be sheetmetal based as that is the tools I have access to, so I'm looking at having the keys plate mounted and having the left and right keyboard mounted on a chair.  As far as curvature goes that isn't really much of a problem for me as I can do that pretty easily, but I've been having trouble trying to determine the best direction for curvature and the dimensions.

Also what have you used to draw the model and where have you sourced the switch models?

(Attachment Link) 47545[/ATTACH]

Interesting, are you going to use a curved PCB or just and wire it?
 I used SolidWorks for the model and I did the switch myself from the pictures and dimensions here.

[bpiphany]

Tin both pads, hold the 0402 down to the board with tweezers hit 1 pad with the soldering iron briefly.  Check to see if the other pad got sufficiently heated to be properly soldered.  Sometimes it happens.  Then finish it out.

Yeah, the video pretty much shows how to do it, nothing new there either really. The thing is though, I could barely grab one of them with my "pointy" tip tweezers. Have to go needle tip with these suckers. I can probably fit ten or so of them on the tip of my soldering iron as well... Have to consider spacing on the layout as well to be able to get down to all solder pads.

Man, thee things are sooo tiny, you should have gone with some 0603 or 0805.

I am going to seriously consider this.

[Icarium]

Tin both pads, hold the 0402 down to the board with tweezers hit 1 pad with the soldering iron briefly.  Check to see if the other pad got sufficiently heated to be properly soldered.  Sometimes it happens.  Then finish it out.

Don't do that. If the tin on the other pad doesn't heat up properly the component will be raised above the pad. Tin one pad, place component and solder in place, then do the other pad.

I have a stereo microscope and some very fine tips. It helps but it doesn't make the shaking go away.

[Findecanor]

As far as curvature goes that isn't really much of a problem for me as I can do that pretty easily, but I've been having trouble trying to determine the best direction for curvature and the dimensions. (Attachment Link) 47545[/ATTACH]

It is not that easy. The edges of each switch hole have to be straight and parallel or the switches will not fit properly.
You would have to separate the columns ...

[jblack801]

I hope it gets easier in the final product. I am comfortable with switch sized leads, but I don't own a microscope or steady hands for components this tiny. hehe

[QuinRiva]

Deflection is pretty small (0.3mm difference in height between the edge and the centre, and 0.2mm in the y-axis, but the later can be accounted for by bend deductions).  However I take your point, so I might try to knock something up with a transverse FR curve along the x-axis.  As far as precision goes, my pressbrake is rated at 0.002mm so that shouldn't be an issue:
[ Attachment Invalid Or Does Not Exist ] 47655[/ATTACH]

[QuinRiva]

I've removed all curvature across the key-holes, and I'm working on the edges.  I've made the corners square for the moment, but I think that I've worked out a way around that, but it uses a crap load more metal.

Dox, is it possible to get a dimensional drawing of your design?

[ Attachment Invalid Or Does Not Exist ] 47771[/ATTACH]

[dorkvader]

QuinRiva: that's a good looking keywell. If we could get kinesis-like keywells on this project, it would be amazing!

Couple of manufactureing difficulties:
1. it would be significantly harder to make. I imagine the only cost-effective way to get that shape would be to stamp it
2. It raises the question on how to make a "base" for it. I wolud be fine with what is shown in your image, having an "open" keyboard like that, but if there needs to be some sort of case made, I imagine it would be difficult
3. Making a PCB for that would be difficult, involving either multiple PCBs with connectors, or some sort of shaped/curved/flexible PCB

Nevertheless, the design is good, and all the issues aren't insurmountable. I imagine one could solder it point-to-point without too much difficulty. I don't know what it'd be to stamp them, though.

[hoggy]

Maltron keyboards are all signed by the person who makes them.  Any plans for the people involved in this project to do the same?

[Input Nirvana]

Like this thread!

I like signing them too...how cool is that??

[QuinRiva]

QuinRiva: that's a good looking keywell. If we could get kinesis-like keywells on this project, it would be amazing!

Couple of manufactureing difficulties:
1. it would be significantly harder to make. I imagine the only cost-effective way to get that shape would be to stamp it
2. It raises the question on how to make a "base" for it. I wolud be fine with what is shown in your image, having an "open" keyboard like that, but if there needs to be some sort of case made, I imagine it would be difficult
3. Making a PCB for that would be difficult, involving either multiple PCBs with connectors, or some sort of shaped/curved/flexible PCB

Nevertheless, the design is good, and all the issues aren't insurmountable. I imagine one could solder it point-to-point without too much difficulty. I don't know what it'd be to stamp them, though.

1.  Not really, as I have stated I intend to use a laser and press-brake to manufacture.  Cost effectiveness is difficult to gauge as the large number of folds, most of which need to be accurate, demand a good-press brake and experienced operator.  I have access to a good pressbrake and my operator is an Astro robot.  However Dox's design could be manufactured on the cheap using sheet metal particularly if he squared the corners.
2.  I've already designed the base (see pictures below); however I don't intend to have an enclosure that sits around the keys.  It wouldn't be too hard to make, I just don't think that it is necessary for plate mounted keyboards.
3.  Not really, I intend to just use a flexible PCB.  They can be a bit tricky to solder on, so I think that the best option is to use a solid PCB for the core components, and then a flexible PCB for the keys.

Hope that makes some sense.  Keep up the suggestions though, as this is my first time at this sort of thing.

[ Attachment Invalid Or Does Not Exist ] 47882[/ATTACH] [ Attachment Invalid Or Does Not Exist ] 47883[/ATTACH] [ Attachment Invalid Or Does Not Exist ] 47884[/ATTACH]

[sordna]

I like the idea of the bend-to-create-a-case, but I don't like the curvature. First of all it's too exaggerated, the pinkies would likely suffer having to hit the edge columns at such an angle. Anyway, I would like this keyboard to be flat for the following reasons:

1. It would make it much easier to type with one hand whenever you wanted to
2. It would make it easier to type with 2-3 fingers on each hand (RSI battling technique)
3. We don't need another Kinesis, whoever wants keywells can buy a pair from Kinesis with red or brown switches soldered in place and ready to go for just $90.

[Icarium]

Right, I was about to say the same thing as Sordna, just a little different. If you want keywells, buy Kinesis keywells and make the rest of the case. Kinesis keywells are plastic but that shouldn't be a problem because there are keys on top of it so you can't see it anyway.

Apart from that I would say, if you're going for keywells make sure they're different from the Kinesis it would be much more interesting.

Personally I am very much looking forward to the flat version as well.

[hoggy]

Custom case for a pair of kinesis keywells sounds like a great spin off project for someone...

[Jim66]

Right, I was about to say the same thing as Sordna, just a little different. If you want keywells, buy Kinesis keywells and make the rest of the case. Kinesis keywells are plastic but that shouldn't be a problem because there are keys on top of it so you can't see it anyway.

Apart from that I would say, if you're going for keywells make sure they're different from the Kinesis it would be much more interesting.

Personally I am very much looking forward to the flat version as well.

I agree with this too.

I much prefer the idea of the original flat project.

[ironman31]

Agreed. The curving keyboard is unappealing to my dollars.

[Findecanor]

I have made a 3D keyboard design that is a middle-ground between Dox's flat keyboard, Kinesis key wells and standard QWERTY. I made several physical models to try it out, both in clay and with key switches and test typed on them. I am still stumped on how to manufacture a small run and not cost a fortune. Anyway ...

One conclusion that I came to is that I don't like the feeling when keys switches are angled too much. The ideal (for me) would actually be if all switches are pointing in the same direction but with curvature being from switches at different heights and with custom key caps.
Slanted columns and traditional key caps may feel better than flat, but only if the curvature is very slight.

[hoggy]

QuinRiva, what I forgot to say is that it looks like a really cool concept.   If it ever goes ahead, please let me know.

Please don't feel under appreciated.

[mbc]

About the cases. You could contact the guy that copied litster's case about producing your cases
 It would be a lot cheaper than aluminium

[QuinRiva]

I wasn't aware that the Kinesis Keywells were available separately, that would substantially simplify the manufacturing process.  Does anyone know how I would go about buying the keywells (with Cherry Reds) and potentially the thumb area from Australia?

Also does anyone know where I could get a dimensionalised drawing of the keywells?

[sordna]

Just go to http://kinesis-ergo.com/ and you'll find contact info. Email them and ask, they sell parts and everything, including keywells with reds which they sell as an upgrade.

[ic07]

Beta firmware's ready .  As far as I can tell from a little testing on my breadboard at least - it looks stable.

The project is on github at https://github.com/benblazak/ergodox-firmware.


The pertinent files at this point are probably:

- The diagram of how i have it set up: [ Attachment Invalid Or Does Not Exist ] 48135[/ATTACH]
(which is also here as an svg, but it looks like you'll have to download it to view it)

- The text documentation for the Teensy 2.0 and MCP23018 code (which is less visually clear, but has lots more notes)

- And the hex file of the current firmware build


For notes on what it can do at the moment see the readme.md (on github) or the last commit message for the master branch.  Let me know what you guys think .

::edit::
Wow, just realized I reversed the row pins for the MCP23018 in my diagram.  It's correct in the .md file though.  I'll have to fix the diagram later.

[Dox]

Awesome job ic07! I'm going to order a few MCP23018 and modify my design accordingly.

Now, what would be the best 4 wire connector/cable.

[dirge]

This threads made me buy teensys and a breadboard.. I think I may be hooked...

[alaricljs]

4 pin Mini-DIN?

4 conductor "stereo" 3.5mm TRRS?

2.5mm TRRS for silly smallness.

RJ11 would be nice but I bet the jack is too big.

[uberben]

You could also go with a standard phone cable. Especially if this is going to be sold in volume, it would allow the end user to replace with whatever length and style (ie curly/straight) they like by running over to the nearest dollar store.

[sordna]

Curly phone cable would be ideal for this application!

[Dox]

A phone cable would be great but unfortunately the 4p4c socket is to high to fit in the case. The same goes for Mini-DIN.

[sordna]

I'm sure there has to be low profile enough sockets... a lot of phones/headsets seem to have pretty compact sockets on them.

[Dox]

I'm sure there has to be low profile enough sockets... a lot of phones/headsets seem to have pretty compact sockets on them.

Sadly no. There is only 11.43mm from the top of the PCB to the top of the case so I would say that the maximum height for the connector is 8mm. The smallest 4p4c I found was about 15mm.

[sordna]

Can't the PCB have a cutaway so that the socket has more room up and down? Of course the socket should be secured by the case, not by the PCB (you would need some flexible leads between socket and PCB).

[Dox]

Can't the PCB have a cutaway so that the socket has more room up and down? Of course the socket should be secured by the case, not by the PCB (you would need some flexible leads between socket and PCB).

No, those sockets need to be soldered/fixed on the PCB and even if I was doing that, the distance from the interior bottom to the top of the case is 16mm (13mm usable for a connector) it's still not enough.
Let's find something else!

[ic07]

I still like the 3.5mm (or 2.5mm) TRRS best, especially aesthetically - as long as it's possible to make it safe from shorting Vcc and any of the other 3.  It looks like the size limitations are going to keep things small enough for my taste whatever gets decided on though .

If using 2 sockets is an option, have you looked at any of those small coaxial power connectors?

Or something simple like a small 4 pin molex?  (Actually, I used to think this connector would be too big and not very pretty, but now I'm starting to think it might look cool...)

sordna:  If you want a curly phone cable anyway, you could always cut the connectors off one and solder some others in their place.  I've been sort of planning on making my own cable if I can, just to get the size and such just right.

[Dox]

I still like the 3.5mm (or 2.5mm) TRRS best, especially aesthetically - as long as it's possible to make it safe from shorting Vcc and any of the other 3.  It looks like the size limitations are going to keep things small enough for my taste whatever gets decided on though .

If using 2 sockets is an option, have you looked at any of those small coaxial power connectors?

Or something simple like a small 4 pin molex?  (Actually, I used to think this connector would be too big and not very pretty, but now I'm starting to think it might look cool...)

sordna:  If you want a curly phone cable anyway, you could always cut the connectors off one and solder some others in their place.  I've been sort of planning on making my own cable if I can, just to get the size and such just right.

Yeah think that the TRRS option is the best for now. Avoiding to short the Vcc is easy, just put the power on the contact that is the furthest outside of the case. That way, there is no chance to short anything.
I prefer to use cables/connectors that are build for external use. That rules out the molex and ribbon cables.
I can't seem to find a pcb mounted 4 contact TRRS 3.5 or 2.5mm. I have to continue searching.

[Parak]

I can't seem to find a pcb mounted 4 contact TRRS 3.5 or 2.5mm. I have to continue searching.

http://search.digikey.com/scripts/DkSearch/dksus.dll?x=0&y=0&lang=en&site=us&KeyWords=CP-43514-ND ?

[ic07]

I can't seem to find a pcb mounted 4 contact TRRS 3.5 or 2.5mm. I have to continue searching.

Would this 2.5mm audio jack work?  Kps linked to it a while ago (I remembered because it was just just before I joined the forum and they were talking about I2C).

Or maybe this 3.5mm audio jack would be better - or maybe there's a different one on there that'd be preferable, I'm not really sure what some of the specs mean so I couldn't tell.

Avoiding to short the Vcc is easy, just put the power on the contact that is the furthest outside of the case.

What about things like if the connector gets plugged in to the right hand side first, and then shorts on the left hand side case (or some keys, or something)?  Or am I just being paranoid now, lol.

[Dox]

Thanks Parak, order placed!

ic07, I prefer the 3.5mm as there is enough place for it and it will be more sturdy.
No matter what side you plug first, with the power on the outside, I think there is no chance to short anything.

[Dox]

ic07, I'm trying to compile your firmware but I get this error.

Code: [Select]



Compiling C: lib-other/pjrc/usb_keyboard/usb_keyboard.c
avr-gcc -c -mmcu=atmega32u4         -I. -gdwarf-2 -DF_CPU=16000000UL -Os -funsigned-char -funsigned-
bitfields -ffunction-sections -fpack-struct -fshort-enums -Wall -Wstrict-prototypes -Wa,-adhlns=./li
b-other/pjrc/usb_keyboard/usb_keyboard.lst -I. -std=gnu99 -MMD -MP -MF .dep/usb_keyboard.o.d lib-oth
er/pjrc/usb_keyboard/usb_keyboard.c -o lib-other/pjrc/usb_keyboard/usb_keyboard.o


Linking: firmware.elf
avr-gcc -mmcu=atmega32u4         -I. -gdwarf-2 -DF_CPU=16000000UL -Os -funsigned-char -funsigned-bit
fields -ffunction-sections -fpack-struct -fshort-enums -Wall -Wstrict-prototypes -Wa,-adhlns=lib-oth
er/pjrc/usb_keyboard/usb_keyboard.o -I. -std=gnu99 -MMD -MP -MF .dep/firmware.elf.d lib-other/pjrc/u
sb_keyboard/usb_keyboard.o --output firmware.elf -Wl,-Map=firmware.map,--cref -Wl,--relax -Wl,--gc-s
ections     -lm
c:/winavr-20100110/bin/../lib/gcc/avr/4.3.3/../../../../avr/lib/avr5/crtm32u4.o:(.init9+0x0): undefi
ned reference to `main'
MAKE[1]: *** [firmware.elf] Error 1
MAKE[1]: Leaving directory `C:/Users/Dox/GIT/benblazak-ergodox-firmware-201e38c/src'
MAKE: *** [all] Error 2

I also tried your compiled .hex and nothing is happening. Any idea what's going on?

[ic07]

I think the compile error was from me using `$(shell find ... '*.c')` in the makefile (which in windows probably meant that no .c files were included in the build at all, lol).  Sorry, I forgot to think about that.  Hopefully it works now (using `$(wildcard ... *.c)`).

As for the .hex, I just downloaded and retested it on my little setup, and it worked...  What should happen is:
- all the LEDs should go high for about 1 second (while USB initializes and we wait for drivers to load)
- then all LEDs should go off unless one of the indicators is currently set (I have numlock on in the picture below)
- it should wait for "keypresses" and generate them if they're detected

Did I do the diagram wrong?  I'll put an actual pic of my setup below just in case.  

breadboard setup:
- note: row and column pins are disconnected; i don't have switches of any type at the moment, so i've been using temporary wires when i needed to test for keypresses
[ Attachment Invalid Or Does Not Exist ] 48226[/ATTACH]

[Dox]

Ok the hex file is working, I had my 2.2k resistors to the ground instead of vcc :doh:.
I can now compile it too with the revised makefile.
Thanks!